4 * Copyright (c) 2006 Junio C Hamano
11 #include "tree-walk.h"
19 * git grep pathspecs are somewhat different from diff-tree pathspecs;
20 * pathname wildcards are allowed.
22 static int pathspec_matches(const char **paths, const char *name)
25 if (!paths || !*paths)
27 namelen = strlen(name);
28 for (i = 0; paths[i]; i++) {
29 const char *match = paths[i];
30 int matchlen = strlen(match);
31 const char *cp, *meta;
34 ((matchlen <= namelen) &&
35 !strncmp(name, match, matchlen) &&
36 (match[matchlen-1] == '/' ||
37 name[matchlen] == '\0' || name[matchlen] == '/')))
39 if (!fnmatch(match, name, 0))
41 if (name[namelen-1] != '/')
44 /* We are being asked if the directory ("name") is worth
47 * Find the longest leading directory name that does
48 * not have metacharacter in the pathspec; the name
49 * we are looking at must overlap with that directory.
51 for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
53 if (ch == '*' || ch == '[' || ch == '?') {
59 meta = cp; /* fully literal */
61 if (namelen <= meta - match) {
62 /* Looking at "Documentation/" and
63 * the pattern says "Documentation/howto/", or
64 * "Documentation/diff*.txt". The name we
65 * have should match prefix.
67 if (!memcmp(match, name, namelen))
72 if (meta - match < namelen) {
73 /* Looking at "Documentation/howto/" and
74 * the pattern says "Documentation/h*";
75 * match up to "Do.../h"; this avoids descending
76 * into "Documentation/technical/".
78 if (!memcmp(match, name, meta - match))
86 static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name, int tree_name_len)
94 data = read_sha1_file(sha1, type, &size);
96 error("'%s': unable to read %s", name, sha1_to_hex(sha1));
99 if (opt->relative && opt->prefix_length) {
100 static char name_buf[PATH_MAX];
102 int name_len = strlen(name) - opt->prefix_length + 1;
105 name += opt->prefix_length;
107 if (ARRAY_SIZE(name_buf) <= name_len)
108 cp = to_free = xmalloc(name_len);
111 memcpy(cp, name, tree_name_len);
112 strcpy(cp + tree_name_len,
113 name + tree_name_len + opt->prefix_length);
117 hit = grep_buffer(opt, name, data, size);
123 static int grep_file(struct grep_opt *opt, const char *filename)
128 if (lstat(filename, &st) < 0) {
131 error("'%s': %s", filename, strerror(errno));
135 return 0; /* empty file -- no grep hit */
136 if (!S_ISREG(st.st_mode))
138 i = open(filename, O_RDONLY);
141 data = xmalloc(st.st_size + 1);
142 if (st.st_size != xread(i, data, st.st_size)) {
143 error("'%s': short read %s", filename, strerror(errno));
149 if (opt->relative && opt->prefix_length)
150 filename += opt->prefix_length;
151 i = grep_buffer(opt, filename, data, st.st_size);
156 static int exec_grep(int argc, const char **argv)
166 execvp("grep", (char **) argv);
169 while (waitpid(pid, &status, 0) < 0) {
174 if (WIFEXITED(status)) {
175 if (!WEXITSTATUS(status))
184 #define push_arg(a) do { \
185 if (nr < MAXARGS) argv[nr++] = (a); \
186 else die("maximum number of args exceeded"); \
189 static int external_grep(struct grep_opt *opt, const char **paths, int cached)
191 int i, nr, argc, hit, len, status;
192 const char *argv[MAXARGS+1];
193 char randarg[ARGBUF];
194 char *argptr = randarg;
197 if (opt->extended || (opt->relative && opt->prefix_length))
207 if (opt->regflags & REG_EXTENDED)
209 if (opt->regflags & REG_ICASE)
211 if (opt->word_regexp)
215 if (opt->unmatch_name_only)
219 if (opt->post_context || opt->pre_context) {
220 if (opt->post_context != opt->pre_context) {
221 if (opt->pre_context) {
223 len += snprintf(argptr, sizeof(randarg)-len,
224 "%u", opt->pre_context);
225 if (sizeof(randarg) <= len)
226 die("maximum length of args exceeded");
230 if (opt->post_context) {
232 len += snprintf(argptr, sizeof(randarg)-len,
233 "%u", opt->post_context);
234 if (sizeof(randarg) <= len)
235 die("maximum length of args exceeded");
242 len += snprintf(argptr, sizeof(randarg)-len,
243 "%u", opt->post_context);
244 if (sizeof(randarg) <= len)
245 die("maximum length of args exceeded");
250 for (p = opt->pattern_list; p; p = p->next) {
252 push_arg(p->pattern);
256 * To make sure we get the header printed out when we want it,
257 * add /dev/null to the paths to grep. This is unnecessary
258 * (and wrong) with "-l" or "-L", which always print out the
261 * GNU grep has "-H", but this is portable.
263 if (!opt->name_only && !opt->unmatch_name_only)
264 push_arg("/dev/null");
268 for (i = 0; i < active_nr; i++) {
269 struct cache_entry *ce = active_cache[i];
271 if (!S_ISREG(ntohl(ce->ce_mode)))
273 if (!pathspec_matches(paths, ce->name))
276 if (name[0] == '-') {
277 int len = ce_namelen(ce);
278 name = xmalloc(len + 3);
279 memcpy(name, "./", 2);
280 memcpy(name + 2, ce->name, len + 1);
283 if (argc < MAXARGS && !ce_stage(ce))
285 status = exec_grep(argc, argv);
292 } while (i < active_nr &&
293 !strcmp(ce->name, active_cache[i]->name));
294 i--; /* compensate for loop control */
298 status = exec_grep(argc, argv);
305 static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
313 * Use the external "grep" command for the case where
314 * we grep through the checked-out files. It tends to
315 * be a lot more optimized
318 hit = external_grep(opt, paths, cached);
324 for (nr = 0; nr < active_nr; nr++) {
325 struct cache_entry *ce = active_cache[nr];
326 if (!S_ISREG(ntohl(ce->ce_mode)))
328 if (!pathspec_matches(paths, ce->name))
333 hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
336 hit |= grep_file(opt, ce->name);
340 } while (nr < active_nr &&
341 !strcmp(ce->name, active_cache[nr]->name));
342 nr--; /* compensate for loop control */
345 free_grep_patterns(opt);
349 static int grep_tree(struct grep_opt *opt, const char **paths,
350 struct tree_desc *tree,
351 const char *tree_name, const char *base)
355 struct name_entry entry;
357 int tn_len = strlen(tree_name);
358 char *path_buf = xmalloc(PATH_MAX + tn_len + 100);
361 tn_len = sprintf(path_buf, "%s:", tree_name);
362 down = path_buf + tn_len;
369 len = strlen(path_buf);
371 while (tree_entry(tree, &entry)) {
372 strcpy(path_buf + len, entry.path);
374 if (S_ISDIR(entry.mode))
375 /* Match "abc/" against pathspec to
376 * decide if we want to descend into "abc"
379 strcpy(path_buf + len + entry.pathlen, "/");
381 if (!pathspec_matches(paths, down))
383 else if (S_ISREG(entry.mode))
384 hit |= grep_sha1(opt, entry.sha1, path_buf, tn_len);
385 else if (S_ISDIR(entry.mode)) {
387 struct tree_desc sub;
389 data = read_sha1_file(entry.sha1, type, &sub.size);
391 die("unable to read tree (%s)",
392 sha1_to_hex(entry.sha1));
394 hit |= grep_tree(opt, paths, &sub, tree_name, down);
401 static int grep_object(struct grep_opt *opt, const char **paths,
402 struct object *obj, const char *name)
404 if (obj->type == OBJ_BLOB)
405 return grep_sha1(opt, obj->sha1, name, 0);
406 if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
407 struct tree_desc tree;
410 data = read_object_with_reference(obj->sha1, tree_type,
413 die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
415 hit = grep_tree(opt, paths, &tree, name, "");
419 die("unable to grep from object of type %s", typename(obj->type));
422 static const char builtin_grep_usage[] =
423 "git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
425 static const char emsg_invalid_context_len[] =
426 "%s: invalid context length argument";
427 static const char emsg_missing_context_len[] =
428 "missing context length argument";
429 static const char emsg_missing_argument[] =
430 "option requires an argument -%s";
432 int cmd_grep(int argc, const char **argv, const char *prefix)
436 int seen_dashdash = 0;
438 struct object_array list = { 0, 0, NULL };
439 const char **paths = NULL;
442 memset(&opt, 0, sizeof(opt));
443 opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
446 opt.pattern_tail = &opt.pattern_list;
447 opt.regflags = REG_NEWLINE;
450 * If there is no -- then the paths must exist in the working
451 * tree. If there is no explicit pattern specified with -e or
452 * -f, we take the first unrecognized non option to be the
453 * pattern, but then what follows it must be zero or more
454 * valid refs up to the -- (if exists), and then existing
455 * paths. If there is an explicit pattern, then the first
456 * unrecognized non option is the beginning of the refs list
457 * that continues up to the -- (if exists), and then paths.
461 const char *arg = argv[1];
463 if (!strcmp("--cached", arg)) {
467 if (!strcmp("-a", arg) ||
468 !strcmp("--text", arg)) {
469 opt.binary = GREP_BINARY_TEXT;
472 if (!strcmp("-i", arg) ||
473 !strcmp("--ignore-case", arg)) {
474 opt.regflags |= REG_ICASE;
477 if (!strcmp("-I", arg)) {
478 opt.binary = GREP_BINARY_NOMATCH;
481 if (!strcmp("-v", arg) ||
482 !strcmp("--invert-match", arg)) {
486 if (!strcmp("-E", arg) ||
487 !strcmp("--extended-regexp", arg)) {
488 opt.regflags |= REG_EXTENDED;
491 if (!strcmp("-F", arg) ||
492 !strcmp("--fixed-strings", arg)) {
496 if (!strcmp("-G", arg) ||
497 !strcmp("--basic-regexp", arg)) {
498 opt.regflags &= ~REG_EXTENDED;
501 if (!strcmp("-n", arg)) {
505 if (!strcmp("-h", arg)) {
509 if (!strcmp("-H", arg)) {
513 if (!strcmp("-l", arg) ||
514 !strcmp("--files-with-matches", arg)) {
518 if (!strcmp("-L", arg) ||
519 !strcmp("--files-without-match", arg)) {
520 opt.unmatch_name_only = 1;
523 if (!strcmp("-c", arg) ||
524 !strcmp("--count", arg)) {
528 if (!strcmp("-w", arg) ||
529 !strcmp("--word-regexp", arg)) {
533 if (!strncmp("-A", arg, 2) ||
534 !strncmp("-B", arg, 2) ||
535 !strncmp("-C", arg, 2) ||
536 (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
540 case 'A': case 'B': case 'C':
543 die(emsg_missing_context_len);
554 if (sscanf(scan, "%u", &num) != 1)
555 die(emsg_invalid_context_len, scan);
558 opt.post_context = num;
562 opt.post_context = num;
564 opt.pre_context = num;
569 if (!strcmp("-f", arg)) {
574 die(emsg_missing_argument, arg);
575 patterns = fopen(argv[1], "r");
577 die("'%s': %s", argv[1], strerror(errno));
578 while (fgets(buf, sizeof(buf), patterns)) {
579 int len = strlen(buf);
580 if (buf[len-1] == '\n')
582 /* ignore empty line like grep does */
585 append_grep_pattern(&opt, xstrdup(buf),
594 if (!strcmp("--not", arg)) {
595 append_grep_pattern(&opt, arg, "command line", 0,
599 if (!strcmp("--and", arg)) {
600 append_grep_pattern(&opt, arg, "command line", 0,
604 if (!strcmp("--or", arg))
605 continue; /* no-op */
606 if (!strcmp("(", arg)) {
607 append_grep_pattern(&opt, arg, "command line", 0,
611 if (!strcmp(")", arg)) {
612 append_grep_pattern(&opt, arg, "command line", 0,
616 if (!strcmp("--all-match", arg)) {
620 if (!strcmp("-e", arg)) {
622 append_grep_pattern(&opt, argv[1],
629 die(emsg_missing_argument, arg);
631 if (!strcmp("--full-name", arg)) {
635 if (!strcmp("--", arg)) {
636 /* later processing wants to have this at argv[1] */
642 usage(builtin_grep_usage);
644 /* First unrecognized non-option token */
645 if (!opt.pattern_list) {
646 append_grep_pattern(&opt, arg, "command line", 0,
651 /* We are looking at the first path or rev;
652 * it is found at argv[1] after leaving the
660 if (!opt.pattern_list)
661 die("no pattern given.");
662 if ((opt.regflags != REG_NEWLINE) && opt.fixed)
663 die("cannot mix --fixed-strings and regexp");
664 compile_grep_patterns(&opt);
666 /* Check revs and then paths */
667 for (i = 1; i < argc; i++) {
668 const char *arg = argv[i];
669 unsigned char sha1[20];
671 if (!get_sha1(arg, sha1)) {
672 struct object *object = parse_object(sha1);
674 die("bad object %s", arg);
675 add_object_array(object, arg, &list);
678 if (!strcmp(arg, "--")) {
685 /* The rest are paths */
686 if (!seen_dashdash) {
688 for (j = i; j < argc; j++)
689 verify_filename(prefix, argv[j]);
693 paths = get_pathspec(prefix, argv + i);
694 if (opt.prefix_length && opt.relative) {
695 /* Make sure we do not get outside of paths */
696 for (i = 0; paths[i]; i++)
697 if (strncmp(prefix, paths[i], opt.prefix_length))
698 die("git-grep: cannot generate relative filenames containing '..'");
702 paths = xcalloc(2, sizeof(const char *));
708 return !grep_cache(&opt, paths, cached);
711 die("both --cached and trees are given.");
713 for (i = 0; i < list.nr; i++) {
714 struct object *real_obj;
715 real_obj = deref_tag(list.objects[i].item, NULL, 0);
716 if (grep_object(&opt, paths, real_obj, list.objects[i].name))
719 free_grep_patterns(&opt);